GraphPPIS is a novel framework for structure-based protein-protein interaction site prediction using deep graph convolutional network, which is able to capture information from high-order spatially neighboring amino acids. The GraphPPIS source code is designed for high-throughput predictions, and does not have the limitation of one query protein per run. We recommend you to use the web server of GraphPPIS if your input is small.
GraphPPIS is developed under Linux environment with:
python 3.7.7
numpy 1.19.1
pandas 1.1.0
torch 1.6.0
To run the full & accurate version of GraphPPIS, you need to install the following three software and download the corresponding databases:
BLAST+ and UniRef90
HH-suite and Uniclust30
DSSP
However, if you use the fast version of GraphPPIS, only DSSP is needed.
- Use
makeblastdbin BLAST+ to build UniRef90 (guide). - Build Uniclust30 following this guide.
- Set path variables
UR90,HHDB,PSIBLAST,HHBLITSandDSSPinGraphPPIS_predict.py.
For a protein chain in PDB:
python GraphPPIS_predict.py -p PDB_ID -c chain_ID
For a user-custom PDB file:
python GraphPPIS_predict.py -f XXX.pdb -c chain_ID
The program uses the fast model in default. If you want to use the slow & accurate mode, type as follows:
python GraphPPIS_predict.py -p PDB_ID -c chain_ID -m slow
We provide the datasets, pre-computed features and the two pre-trained models here for those interested in reproducing our paper.
The datasets used in this study (Train_335, Test_60, Test_315 and UBtest_31) are stored in ./Dataset in fasta format.
The distance maps(L * L) and normalized feature matrixes PSSM(L * 20), HMM(L * 20) and DSSP(L * 14) are stored in ./Feature in numpy format.
The pre-trained GraphPPIS full model and the simplified version using BLOSUM62 can be found under ./Model
The GraphPPIS web server is freely available at https://biomed.nscc-gz.cn/apps/GraphPPIS
Contact:
Qianmu Yuan (yuanqm3@mail2.sysu.edu.cn)
Yuedong Yang (yangyd25@mail.sysu.edu.cn)